On Saturday 15 March 2008 13:40:13 Peter Gibbs wrote: > I thought the same thing, but decided some sort of defensive check was > required anyway, and I got lost trying to track it further. It served my > purpose at the time, which was to resolve a very fragile bug that moved > or disappeared while trying to pin it down. > If you want to track the root cause, just replace the new lines by > PARROT_ASSERT(i+1 < st->dest.n), which seems like a reasonable defense > to leave there.
Fine by me either way. > Incidentally, I found the following useful to stop valgrind complaining > about uninitialized values causes by walking the stack: Very nice. Do we need to compile with -DVALGRIND, or is there special magic somewhere? -- c
